Journal of Physical Chemistry B, Vol.113, No.18, 6533-6542, 2009
Designing an Appropriate Computational Model for DNA Nucleoside Hydrolysis: A Case Study of 2'-Deoxyuridine
This study uses a variety of computational models and detailed, systematic potential energy surface scans to examine the hydrolysis of 2'-deoxyuridine. First, the unimolecular cleavage was studied using a model that only includes the nucleoside. Although comparison of experimental and (PCM-B3LYP/6-31+G(d)) calculated (Gibbs energy) barriers confirms that hydrolysis occurs via a fully dissociative (S(N)1) mechanism with a rate-limiting step of glycosidic bond dissociation, this model does not provide a complete picture of the hydrolysis mechanism. When the model is expanded to include one explicit water nucleophile, gas-phase optimizations are unable to model charge separation in the reaction intermediate, while optimizations that implicitly incorporate the effects of bulk solvent do not accurately model the second reaction step (nucleophilic attack following dissociation) due to insufficient (water) nucleophile activation and (uracil anion) leaving group stabilization. Further expansion of the model to include three explicit water molecules allows for. discrete proton transfer from the water nucleophile to the uracil anion, and thereby generates smooth reaction surfaces for both the dissociative (S(N)1) and concerted (S(N)2) pathways. Furthermore, for the first time, this computational model for the uncatalyzed hydrolysis of the N-glycosidic bond in a nucleoside predicts that the dissociative mechanism is more favorable than the concerted pathway, which supports experimental findings. It is also found that although (implicit) solvent-phase single-point calculations on gas-phase geometries can yield similar energies to solvent-phase optimizations, the geometries can be very different and not all potential reaction routes can be fully characterized. Therefore, care must be taken when interpreting mechanistic information obtained from gas-phase structures. This work provides a template for generating other nucleoside or nucleotide hydrolysis models including those relevant to both uncatalyzed and enzyme-catalyzed reactions.